Fiducial Method For Estimation And Lower Confidence Limit Of Complex System Reliability

Several subsystems in a certain arrangement constitute a complex system. Thesubsystems are usually independent of each other. In terms of reliability, the systemis directly restricted by subsystems. However, actually there is few data informationdirectly from the system, while generally relevant data and information of thesubsystem is easy to obtain. Therefore, using the data of subsysterms to study thereliability of complex systems have always been concerned by the scholars ofdifferent areas.At present, the researches for system mostly have concentrated in some specialsystems. For the complex system described by the minimum path matrix, thesubsystems are independent and life distribution is unkowm. The researches onestimation and lower confidence limit of complex system reliability are not toomuch. The thesis has given the studing results for the following several aspects:1. In the case of complete data from subsystem, according to the sample data ofeach subsystem, the distribution of β(R|s, f), LΓ(R|z,η), has fitted respectivelythe accuracy fiducial distribution f(r) of the system reliability, then giving twoforms of the system reliability lower confidence limit. For the bridge system, thisthesis compared this method with WCF method and asymptotic normal method bysimulation, and the result indicates: this method is better than the normal method,although slightly behind WCF, but relatively simple and easy to use. At the sametime, this thesis also gives the system reliability estimation and its asymptoticdistribution.2. In the case of interval type subsystem data from subsystem, there is currentlyno result for system reliability lower confidence limit. In this case, in this thesisusing the distribution of β(R|s, f), LΓ(R|z,η), have fitted respectively theaccuracy fiducial distribution f(r) of the system reliability, then giving two forms of the system reliability lower confidence limit. Besids comparing serial system withasymptotic normal method by simulation, the result havs denominated that themetod in the paper is more applicable. At the same time this thesis also has given theestimation of system reliability.